/SBN_NVIL

Neural Variational Inference and Learning for Sigmoid Belief Network

Primary LanguageMatlab

Neural Variational Inference and Learning for Sigmoid Belief Network

The Matlab Code for the NVIL inference algorithm for SBN.

Zhe Gan (zhe.gan@duke.edu), 9.22.2015

Introduction

This code implements the Neural Variational Inference and Learning (NVIL) algorithm for Sigmoid Belief Network(SBN). The NVIL was proposed by Andriy Mnih and Karol Gregor in their ICML paper "Neural Variational Inference and Learning in Belief Networks". Our NIPS 2015 paper "Temporal Sigmoid Belief Network" also uses the same algorithm, but for a different model, i.e. a SBN-based time-series model. See the repo zhegan27/TSBN_code_NIPS2015 for details.

License

Please note that this code should be used at your own risk. There is no implied guarantee that it will not do anything stupid. Permission is granted to use and modify the code.

Citing TSBN

Please cite the original NVIL paper and also our NIPS paper in your publications if it helps your research:

@inproceedings{TSBN_NIPS2015,
  Author = {Z. Gan, C. Li, R. Henao, D. Carlson, and L. Carin},
  Title = {Deep Temporal Sigmoid Belief Networks for Sequence Modeling},
  booktitle={NIPS},
  Year  = {2015}
}